pearu / pyvtk Goto Github PK

What steps will reproduce the problem?
1. Write a mesh ( in my case unstructured) with 'float' mesh points
2. Output VTK file will state that the data is still of type 'int'

Looking at the source I see that 'int' is hard coded in UnstructuredGrid.py in 
line 100

I recommend placing a switch that detects whether the mesh coordinates are of 
type 'int' or 'float' then write the correct string.

I've attached a sample VTK written by pyvtk where the issue may be seen (I left 
it in ASCII format). The error is on line 5 of the sample newTest.vtk

Original issue reported on code.google.com by [email protected] on 25 Aug 2012 at 9:48

Attachments:

• newTest.vtk

Thanks for creating this very useful package.

I am trying to learn to use it by following the examples. When running unstructured.py, I get the following error

      1 vtk = pyvtk.VtkData(  pyvtk.UnstructuredGrid(points,
----> 2     triangle=tri.simplices
      3     ),
      4   pyvtk.PointData(pyvtk.Scalars(pointPressure,name='Pressure')),
      5   pyvtk.CellData(pyvtk.Scalars(cellTemp,name='Temperature')),

/home/rpa/.conda/envs/lagrangian_vorticity/lib/python2.7/site-packages/pyvtk/UnstructuredGrid.pyc in __init__(self, points, vertex, poly_vertex, line, poly_line, triangle, triangle_strip, polygon, pixel, quad, tetra, voxel, hexahedron, wedge, pyramid, quadratic_tetra)
     71                 self.vertex = r
     72             if self._check_int_seq(getattr(self,k),sz):
---> 73                 raise ValueError('In cell %s: must be (seq of seq|seq) integers less than %s'%(k,sz))
     74 
     75         for k,n in self._vtk_cell_nums_map.items():

ValueError: In cell triangle: must be (seq of seq|seq) integers less than 1000

I am not very familiar with the internals of the package, so it's hard for me to debug what is going on.

Dear developers,
I found a bug when exporting tensors in Python 3. There is apparently missing conversion from string to bytes. The solution is to fix function Tensors.py::def to_string as

def to_string(self,format='ascii'):
t = self.get_datatype(self.tensors)
ret = [('TENSORS %s %s'%(self.name,t)).encode(),
self.seq_to_string(self.tensors,format,t)]
return b'\n'.join(ret)

This is similar logics as in Scalars.py or Vectors.py at the same functions def to_string().
Regards Vit

Hey,

In UnstructuredGrid.py, as for an example, there is a function unstructured_grid_fromfile, the first line calls a .decode() that won't work with Python3, since str are not decode-able.

The Python3 support seems not totally functionnal for now.

A patch would be to use unicode_literals from future in both python2 and 3

I'd like to make PyVTK available through conda-forge, and it would be helpful (maybe required) to have a LICENSE for pyvtk.

I can offer a pull request, maybe with a BSD 3 license, or any other license the author (@pearu) would prefer.

Many thanks,

Hans

What steps will reproduce the problem?
1. import pyvtk
2. data = pyvtk.VtkData('data.vtk') # This is a binary structured_points dataset

I see that this raises a NotImplemented error.  I am mostly curious to know if 
there are plans to support this in the near term.  I couldn't find a mailing 
list so hopefully this will reach the right people.  

If there aren't plans to implement it, would it be possible to point me in the 
right direction and I'd be happy to take a swing at it.

Thanks,
Sam Skillman

Original issue reported on code.google.com by SamSkillman on 13 Aug 2012 at 4:44

The page https://github.com/pearu/pyvtk/wiki/unexpectedEOF suggests a way to fix UnexpectedEOF on files which are conforming to the ambiguous standard for VTK2. 3rd-party projects using pyvtk have to monkey-patch pyvtk to open such files (see https://sourceforge.net/p/mupif/code/ci/master/tree/mupif/VtkReader2.py at the end of the file for a case in point).

This workaround should be either optionally enabled in pyvtk (without monkey-patching), or better yet (as it is safe as far as reading valid input is concerned) change the split(' ') to split() for good so that both variants are readable.

Thanks for a very useful package.

It might be that I dont understand VTK correctly but I cannot get pyvtk to write results as a time series in a single vtk-file. I have a number of CELL_DATA fields which I would like to see in Paraview as a time series. The only way I can currently make this work is by writing out one vtk-file per field, ie

model_with_field_0.vtk
model_with_field_1.vtk
model_with_field_2.vtk

This duplicates my UnstructuredGrid data which I think is unnecessary (?). Is there a way in pyvtk to get multiple CELL_DATA fields into a single vtk file, or is this not possible in legacy vtk file format? Or is there a way to split UnstructuredGrid-data in one file and then CELL_DATA in separate files?

Thank you and sorry if this is the wrong forum to ask this.